I've never developed so fast!! (And it's only getting stronger!)
Machine translation from the original post. Not reviewed by a human translator. Historical claims may no longer be current.
I've never developed so fast!! (And it's only getting stronger!)
10:00 Cursor open on an empty folder
10:05 Finished writing the prompt (yes, I invested at least 5 minutes in it, that's the secret)
10:30 (while I'm busy with other things) Cursor finished building me a browser extension to improve the GCP interface
After it searched the web and researched how best to technologically build the extension according to my instructions, checked itself, built the extension, and finished by explaining to me which folder holds the code I need to load.
I spent the next few iterations fine-tuning the design (only because I didn't invest those minutes in the first prompt) and two more prompts got me to a result so impressive that I would have paid for such a product!
It all started with a production incident in which someoneee, no names, I held back from tagging you know who you are, pasted the wrong JSON into Secret Manager and crashed production (yes, that's a different conversation about how to manage it, let's not stray from the topic) but it made me wonder how it is that Google, the company that rules this planet, has such an appalling interface and literally no one out of the million employees it has has done anything about it to this day?
So I put on the cape of the savior of humanity from lousy interfaces again and equipped myself with the best AI I could find and... it worked like magic! Even I was surprised that I didn't need more than one iteration! Just a good prompt and a good model (Opus, you're a lifesaver, you should know, nothing less than a quiet revolution)
The more you work with AI, the more you understand that it's not (only) a code-writing engine, it's a real technological partner for consultations and research.
My prompt started with the words "I have a problem"
After describing the problem, and pouring out amounts of information - everything I could put in so that it would have as much context as possible (links to our projects, our guides whose content I copied from Confluence, past attempts and similar projects)
I also shared how I think it should be solved (a browser extension maybe?)
But I didn't ask it to do what I think, I asked it to research on its own and propose a plan
And when I say research, I really insist that it take the time to learn and investigate and turn over every stone so that it has as full a picture as possible of how to implement everything end to end
Only after that did we set off.
To think that less than 3 years ago I sat down to build, alone, a browser extension that improves the Jira interface (an extension that was also published on Geektime for those who remember) and I sweated a few weekends over the code to make that thing work
And today I create a solution like this between meetings
And in a technology I haven't had a chance to work with before (Plasmo)
It's a joy to live in an era where the technological creativity barrier has been lifted and you can simply create with a free hand and solve small troubles and big ones
I attached a video that demonstrates the extension for the curious who read this far...
Want to see the prompt I wrote too? 😇